[9fans] vacfs

[9fans] vacfs

[ron minnich]

OK, I am trying vacfs on a good system (via vmware) and ... it doesn't
work there either. So that explains part of the problem.

I use the scripts in /sys/src/cmd/venti to dump some scores, take the text
score and dump it into a file in binary, verify with xd that it really
looks right (it would get a different error if it were a bad score, I
tested this) and run vacfs and ... get the vfsOpen: fatal error: short
read

The venti servers do print some output but vacfs itself just can't get
anywhere. oh ho. anybody out there used vacfs lately :-)


ron

Re: [9fans] vacfs

[ron minnich]

On Fri, 6 Feb 2004, ron minnich wrote:

> I use the scripts in /sys/src/cmd/venti to dump some scores, take the text
> score and dump it into a file in binary, verify with xd that it really
> looks right (it would get a different error if it were a bad score, I
> tested this) and run vacfs and ... get the vfsOpen: fatal error: short
> read

good old acid -l truss. Oops. Looks like I misread the source: vacfs is
trying to read 44 bytes, which I assume is non-binary. foolish me!

OK, back to work :-)

ron

Re: [9fans] vacfs

[ron minnich]

Use a file that looks like this: vac:<20 hex digits> and it all works
fine. Cool. Turns out the 'short read' message, which I thought meant
short read from the venti server, actually meant that the score file was
too short. Possibly the vtSetError("short read") code in
/sys/src/cmd/vac/fs.c could change to
vtSetError("vac file: too short, should be >= 44 bytes"),
which would be more useful to confused folks like
me.

I realize this is catering to the clueless, which only encourages them,
but what the heck, I'm one of them!

Now back to the real Plan 9 system. Lights at end of tunnel?

ron

p.s. I hope nobody minds this running soap story, but I figure somebody
else will hit this prolem someday, and I'm hoping they'll find this thread
in the archives.
p.p.s. It's impressive as all hell that ALL the source code for a running
Plan 9 system is on the boot CD!

Re: [9fans] vacfs

[boyd, rounin]

> I realize this is catering to the clueless, which only encourages them,
> but what the heck, I'm one of them!

nope, a precise, terse, informative error message is invaluable.

Re: [9fans] vacfs

[ron minnich]

OK, vacfs is working fine on the node with the dead fossil and the still
valid venti arenas.

Here is what is neat: I had 2 arenas and 2 indexes, which I laid out this
way by following the wiki and man pages. The CDROM venti was very unhappy
with the 2 indexes, so I took a chance, changed the config file to only
have one index (/dev/sdC0/isect0), and rebuilt the index with buildindex.

All this from the CD. All the source available for perusal, so confusion
could be reduced. Any time things got weird, slay venti and start over.

Then I sucked the good score from the dead fossil, and using that score I
have been able to use vacfs to get my last venti mounted via vacfs.

Venti is very impressive, and these tools are really pretty slick.

Looks like I can recover my files, which is a real relief.

One thing that will always be useful is the FAT partition. It's just a
handy place to have around in the event of disaster and you have to boot
from CD. I think I'm going to make it bigger in future just to have it
there. Sure you have ramfs but if you have to reboot you don't want to
lose any temp files you've worked up.

ron

Re: [9fans] vacfs

[boyd, rounin]

> One thing that will always be useful is the FAT partition.

yup, always have two paths [1 level of redundancy/fallback].

Re: [9fans] vacfs

[vdharani]

> OK, I am trying vacfs on a good system (via vmware) and ... it doesn't
> work there either. So that explains part of the problem.
>
> I use the scripts in /sys/src/cmd/venti to dump some scores, take the
> text score and dump it into a file in binary, verify with xd that it
> really looks right (it would get a different error if it were a bad
> score, I tested this) and run vacfs and ... get the vfsOpen: fatal
> error: short read
>
> The venti servers do print some output but vacfs itself just can't get
> anywhere. oh ho. anybody out there used vacfs lately :-)

not sure if it helps but i too have had problems with vacfs (particularly
when i tried multiple vacfs mounts in series). i think vacfs worked after
a venti restart.

regards
dharani

Re: [9fans] vacfs

[Joel Salomon]

> One thing that will always be useful is the FAT partition.

Of course with Micro$oft pulling an SCO....
http://www.microsoft.com/mscorp/ip/tech/fat.asp

--Joel

Re: [9fans] vacfs

[ron minnich]

On Sat, 7 Feb 2004, Joel Salomon wrote:

> Of course with Micro$oft pulling an SCO....
> http://www.microsoft.com/mscorp/ip/tech/fat.asp

we'll just stick to fat16 :-)

ron

Re: [9fans] vacfs

[Charles Forsyth]

[-- Attachment #1: Type: text/plain, Size: 86 bytes --]

just wait until they work their way round to the `start button'/`toolbar' patent(s).

[-- Attachment #2: Type: message/rfc822, Size: 2439 bytes --]

From: ron minnich <rminnich@lanl.gov>
To: 9fans@cse.psu.edu
Subject: Re: [9fans] vacfs
Date: Sun, 8 Feb 2004 09:32:41 -0700 (MST)
Message-ID: <Pine.LNX.4.44.0402080932290.13641-100000@maxroach.lanl.gov>

On Sat, 7 Feb 2004, Joel Salomon wrote:

> Of course with Micro$oft pulling an SCO....
> http://www.microsoft.com/mscorp/ip/tech/fat.asp

we'll just stick to fat16 :-)

ron

Re: [9fans] vacfs

[a]

// we'll just stick to fat16 :-)

bah, random-access is overrated. just put a tar file on
raw media and append. ;-)
ア

Re: [9fans] vacfs

[Wes Kussmaul]

That would give the courts reason to look again at Windows' source of
inspiration: Xerox Alto and the Mac

>just wait until they work their way round to the `start button'/`toolbar'
>patent(s).
>
> > Of course with Micro$oft pulling an SCO....
> > http://www.microsoft.com/mscorp/ip/tech/fat.asp
>
>we'll just stick to fat16 :-)
>
>ron

Re: [9fans] vacfs

[rog]

it'd be nice if programs that expect a vac score would accept
a direct vac:4c7aacf3... score on the command line.

i always forget and then have to improvise with:

	vacfs <{echo vac:4c7aacf3...}

which seems a bit unnecessary.

or is there a reason for this restriction? if not, i'll submit a change.